DOCUMENT RESUME ED 355 807 AUTHOR Basabas-Ikeguchi ... · DOCUMENT RESUME ED 355 807 FL 021 072...
Transcript of DOCUMENT RESUME ED 355 807 AUTHOR Basabas-Ikeguchi ... · DOCUMENT RESUME ED 355 807 FL 021 072...
-
DOCUMENT RESUME
ED 355 807 FL 021 072
AUTHOR Basabas-Ikeguchi, CeciliaTITLE Analysis of Reading and Listening Comprehension
Skills in Different Language Environments.PUB DATE 88NOTE 32p.; Master's Thesis, Dokkyo University (Japan).PUB TYPE Reports Research/Technical (143)
Dissertations /Theses Masters Theses (042)
EDRS PRICE MF01/PCO2 Plus Postage.DESCRIPTORS *Cloze Procedure; Comparative Analysis; *English
(Second Language); Foreign Countries; High Schools;High School Students; *Language Proficiency; LanguageTests; *Linguistic Theory; *Listening Comprehension;*Reading Cnmprehension; Second Language Learning
IDENTIFIERS Unitary Theory
ABSTRACTThe unitary competence theory claims the existence of
a general language proficiency that makes the global evaluation oflanguage skills possible and accounts for the high correlationsbetween language tests of whatever form and modality. This study aimsto test the theory by investigating the correlation between readingand listening comprehension of students under different linguisticenvironments. The cloze method was used to measure the readingcomprehension of students in a second and foreign language setting,as well as to predict the degree of development of these languageskills. The study also investigated whether differences in exposureto English as a Second Language (ESL) and English as a ForeignLanguage (EFL) yield differences in the degree of development of thelanguage skills, and prove which of the two kinds of exposures leadsto greater or lesser skills differentiation. The low correlationbetween the test rests leads to a question of the unitary competencetheory. The results suggest that the cloze system may be a predictorof the differentiated degree of language skills development. Of thetwo groups taken as samples, the performance of the ESL students inthe tests reveal that there is greater tendency for the listeningskills to be specialized more than reading comprehension, or viceversa, while exposure to EFL may lead to less differentiation.Contains 37 references. (Author/LB)
***********************************************************************
Reproductions supplied by EDRS are the best that can be madefrom the original document.
***********************************************************************
-
toO
RC & LC Skills Devlpt. in a2
TITLE PAGE
ANALYSIS OF READING AND LISTENING
COMPREHENSION SKILLS IN DIFFERENT
LANGUAGE ENVIRONMENTS
Dr. Cecilia Basabas Ikeguchi
Faculty of Foreign Languages
Dokkyo University, Japan
"PERMISSION TO REPRODUCE THIS
MATERIAL HAS BEEN GRANTED BY
Clack sci.too.v ttcL;
TO THE EDUCATIONAL RESOURCES
INFORMATION CENTER (ERIC)."
BEST COPY AMBLE
2
U.S. DEPARTMENT OF EDUCATIONOffice of Educational Research and Improvement
EDUCATIONAL RESOURCES INFORMATIONCENTER !ERIC)
Ilieft:s document has been reproduced Sireceived from the demon or ofganozabonoriginating
0 Minor changes have been made to mono.*TeDrOdubtion quality
. intS of view Or opinions stated in thisdocu
ins .1 40 not necessarily represent oNcual
OERI position or pOSCy
-
RC & LC Skills Devlpt. in El 1
Reading and Listening Comprehension. Skills Development
in Different Linguistic Environments
- in Relation to the Clone -
Two competing theories have dominated the field of language testing
research which seek to demonstrate and explain the structure of language
proficiency and measure proficiency levels using external validating criteria.
Empirical evidences support the claim of two extreme alternatives: at one end
of the theoretical continuum is the claim that the language learner seems to
acquire an "overall proficiency" which operates in all the language skills, and
this unitary competency leads to the hypothetical possibility of a single unitary
factor that accounts for all the reliable variance in all tests of "whatever mode,
whatever form". (Oiler 1978, 1980, 1983, 1985) The existence of a "general
language proficiency" is evidenced by the generally high correlations among a
large variety of language competence variables. At the other end of the
pendulum is the divisible competence hypothesis that argues that language
proficiency can be divided into separate components or several aspects of
them. The proponents of the latter claim the possibility of an accountable
number of distinct elements of language competencies which prevent any sort
of global evaluation. (Lado 1979, 1986) Somewhere in between are the
arguements raised by Carroll that there is a "general language ability" , but at
the same time, that language skills have some tendency to be developed and
-
RC & LC Skills Devlpt. in El, 2
specialized to some degrees, or at different rates, so that different language
skills can be separately recognized and measured (1983). Carroll advocates a
compromise between the two extremes, preferring a solution that allows a
general factor plus various more specialized factors associated with particular
skills, components of knowledge and the like.
This study is concerned with the extent to which the unitary competence
theory has been explored in ESL and EFL study in relation to the doze
method which has been used to measure the English reading comprehension
of none-native English learners. There is wide support from linguistics as
well as from language testing research on the use of doze procedure as a
measure of reading comprehension. Its advocates, Oiler in particular, claim
that this test type is one of the integrative tests (1983) which measure global
skills and correlate highly with other integrative tests, especially that of listening
comprehension. The most important arguement in its favor is that it require..;
the FL or SL student to perform a task which is not unlike that which native
speakers do in sending and receiving messages.
Recent researches on the use of the doze test have focussed on the
question of construct validity: they have examined the extent to which doze
deletions are capable of measuring comprehension through language abilities
beyond the knowledge of the sentence-level grammatical structure. On the
one hand there are those who argue that doze tests are not sensitive to context
beyond the sentence. Alderson (1980) systematically varied the size of fixed-
-
4
RC & LC Sldlls Devlpt. in El: 3
deletion ratios across passages of differing difficulty and across groups, and
found no evidence that increase in context affected the subjects' ability to
complete the items successfully. Shanahan, Kamil and Tobin (1982) corstrW.,ed
fixed-ratio doze tests from intact passages, randomly scrambled pages and
passages in which the original sentences were extracted and inserted into
unrelated texts. They found no significant difference in their subjects'
performance across these three conditions. The conclusion reached was that
doze tests do not measure the integration of intersentential information
integration, and it seems to be unreasonable to use and interpret doze in the
classroom as a global measure of reading comprehension. Similarly Alderson
and Porter (1983) report an empirical finding that doze tests are either only
negligible or else not at all sensitive to constraints ranging beyond 5-10 words
of context on either side of the blank. Charles Alderson poses a number of
penetrating questions about the validity of doze tests. He argues that all doze
tests measure the same skill or complex of skills and that in general doze
tests may not be as valid as what it hadbeen previously claimed to be. Aldersons
findings claim that "This is not to assert that doze tests are incapable of
testing more than the comprehension of the immediate environment, but that
as a test, the doze is largely confined to the immediate environment of a
blank. ..." Moreover the fact that the high correlation of the semantically
acceptable scoring method with measures of ESL proficiency, which is designed
to be relatively insensitive to long-range contextual constraint, correlated highly
with the exact -scoring procedure, appears to add support to the thesis that
doze is essentially Clause-bound.
-
RC & LC Skills Devlpt. in H2 4
On the other hand, another group of researches have found support
for the claim that doze items are capable of measuring textual relationship
beyond the level of sentences. Chihara, Oiler, Weaver and Chavez -Oiler (1977)
using a fixed-ratio deletion procedure with intact and scrambled texts found
that both native and non-native speakers of English performed much better on
doze items in intact texts than they did on the same items in the scrambled
texts. In addition they found that the percentage of closures was higher for
groups with higher language proficiency (native speakers and advanced ER,
students) than for groups of lower proficiency (beginning and intermediate
EFL students). Bachman (1982) using a rational deletion procedure with an
intact passage, found that deletion types requiring context beyond the clause
for closure loaded on different factors rather than did deletions which depended
only upon the context of the clause in which they occurred. The arguements
raised in the study are based on the assumption that within a given text more
words function syntactically than cohesively, a random deletion procedure
would tend to sample a large portion of clause-bound words, and therefore
appear to be measuring only lower-level skills. But the findings that the
high-level proficiency non-native speakers performed very similarly to native
English speakers on both syntactic and cohesive doze items, while low
proficiency English sepakers performed significantly poorer on cohesive items,
suggest that doze passages using a rational deletion procedure can be used to
measure textual relationship beyond clause boundaries. More recently the
experiment by Chavez - Oiler, Chihara and Weaver and Oiler (1985), using
-
k.
RC & LC Skills Devlpt in El, 5
previous data (1977) indicates that the performance of the subjects (both
native and non-native English speakers) differed mostly between the scrambled
and syntactic conditions. A linguistic analysis demonstrated that virtually all
the items depended on intersentential context forsuccessful closure.
The performance of EL1 and EL2 respondents onE3achmarfs innovative
experiment with several deletion types (1985) further throws light on the
question of whether doze measures higher orderskills. Bachman, in this
study, was primarily concerned with constructing vaned and reliable doze
tests for reading development; however, the different responses of the EL1
and Fl 2 subjects on the different deletion types support the claim that doze
items are sensitive to discourse across sentences. These variations in
performance of native speakers as well as thoseof the non-native speakers
might indicate that doze tests do measure higher-orderskills.
Along with the external component of constructvalidation, a substantial
amount of research on doze items hasfound evidence for some daze items as
measures of textual competence (Shanahan and Kamil 1982) and others as
measures of grammatical competence. (Weaver and Oiler 1985, Lado 1986
and Markham 1985). Much of the research on doze procedure mentioned
above consisted of the analyses of doze and other language tests which were
all shown to correlate to some degree (Shikkani 1986) or to load in one
general factor (Oiler, 1983). A single factor which accounts for most of the
variance in language tests has been reported in many studies in which doze
i
)
-
RC & LC Skills Devil% in El, 6
has been used. (Oiler and Hinofitis, 1980) and Alderson (1950,1983) Although
Alderson does not refer to a general factor, the assumption that doze measures
a rather homogenous group of lower-order core proficiency skills seems to
underlie much of his research.
The present experiment was conducted to partially determine the
discriminative power of the doze test (scored by the exact and the acceptable
scoring methods) and its ability to predict other language skills, specifically
that of listening comprehension. The basic concern of this study is to
determine whether differences in the degree of exposure to 112 - as a foreign
language and as a second language - would result in differences in the degree
of development between the language skills, particularly those of listening and
reading almprehension.
Experimental Method
Subjecti
A total of 230 subjects, consisting of two groups of high school
students, were taken for the experiment. The first (EFL) group consisted of
116 first year senior high school students in Japan; the second (ESL) group
consisted of 114 third year high school students in Manila. The age range of
the population was from 15-16. Since the students in group 1 were enrolled in
a high school course where no attempt had been made at ability grouping, it is
assumed that they were representative of the larger population from which the
samples were drawn. Meanwhile the students in the second group were selected
-
RC & LC dlls Devlpt. in B.2 7
from one high-ability class, one average-ability class, and one low-ability
class, as per the structure of the school they are enrolled in.
It is at this point where a careful word of distinction is necessary to
explain what may be a verbal quibble concerning the linguistic environments
of the subjects of this paper. English as a Second Language and English as a
Foreign Language. By English as a Foreign Language for the (hereon referred
to as the ) first group of Ss, is meant that English is taught as a school
subject or on an adult level solely for the purpose of giving the students a
foreign language competence which he may use in one of several ways - to
read literature, to read technical work, to listen to the radio, to understand a
dialogue in the movies, to use the language for communication possibly with
transient English or Americans. In general the students' motivation is for the
practical use of English, while some of them realize at the same time that the
language is an effective key to the understanding of another culture.
The term English as a Second Language is used to describe the situation
of the (hereon referred to as) second group of Ss where English becomes a
language of instruction in schools. It is often fundamental to the life and
functioning of the country rather than a region where it has a foreign language
status. But another word of caution has to be pointed out here: it is that in
situations involving both English as a Second or Foreign Language, the word
2egappears to be an understatement In the case of the language environment
of the second grouii of Ss, the school child has had contact with English as
-
F
RC & LC Skills Devlpt in E 8
early as right after birth, but the first language comes first - or simultaneously
- with another regional dialect This means that English learning is complicated
by interference from two sets of language structures instead of just one. So
even between English learners in an L2 setting, there are differences between
learning motivations, type and amount of exposure to the target language, as
well as points of emphasis on the L2 learning at home and in school, as much
as there are critical differences between first and second language learning as
discussed by Elley and Mangubhai (1983).
Experimental Materials and Design
The doze passage used in the research was adopted from the text 1-ligh
Road to English, a high school reader. The selected passage contained
approximately 350 words, the title of which is "Four Boys and the French
Cave of Lascaux ". The close itself was constructed by deleting every seventh
word in the passage for a total of 50 blanks. Two sentences were left unmutilated
at the beginning of the passage and one at the end to provide a complete
context
The passage had a Flesch readability (1948) score of 9, ranking it
as "easy" and a human interest score of 50, indicating that the article is
"highly interesting". The Dale-Chall formula (1948) ranked it as appropriate
for the fourth or fifth grade, with a score of 4.8. This indicates the reading
grade of the student who could answer one-half of the test questions correctly.
-
RC & LC Skills Devlpt. in El2 9
Both the SMOG and the Fry formulas gave approximately the same readability
grade for the passage.
For the listening comprehension test, a TOFEL exam used in a Japanese
school was used with the Ss. Under the present conditions, it was assumed
that a TOFEL test would be a more valid measure for both the Ss ratherthan a
TESOL exam which would be more difficult for the first group of samples.The
listening comprehension test totalling 65 points consisted of seven parts. In
the first part, the Ss were asked to select one logical answer from among three
choices given. In the second part, the students heard ten short questions that
follow a recorded conversation. Then they were asked to select one logical
answer from among the three choices given. Part Three was understanding of
three recorded passages which were played twice. The students were instructed
to answer questions relevant to the passages by choosingfrom choices of
three items each. In Part four, the Ss were asked to decide if the meaning
expressed by each statement is logical or not Part five consisted of a much
longer passage than those in Part three. The students heard the passage twice,
after which they head five short sentences. They were asked to decide whether
the sentences they heard were relevant to the passage or not Part six consisted
of ten recorded short questions which the Ss heard only once. They were then
instructed to select one appropriate answer from three given choices. Part
seven consisted of question-answer form for dal.), conversation. The students
were asked to select from three choices the answer that best fits the given
situation.
The subjects were tested in groups of 40 persons or more according to
-
0
RC & LC Skills Devlpt in Ell 10
the class size. Each of the subjects took the text of the close passage on the
first day and the listening comprehension test on the second day. The tests
were administered in a regular class time under regular classroom conditions.
For the first group of students, instructions for the doze test were read
aloud in English and then explained in Japanese by the Japanese examiner.
For the second group, instructions were first read aloud by the researcher and
were read silently by the students themselves. They were instructed to read
over the passage quickly for general meaning, then to fill in the blanks, and
finally to check their answers. The LCT was recorded on tape by a native
speaker, and the students were allowed to listen to each item of the test, and
answer the questions within the time limit indicated in the tape. The subjects
were given 30 minutes to answer the doze test and a fourty-minute class
period to answer the listening comprehension test.
Results and Discussion
Scoring
Researches with native English learners have explored the various
methods of scoring doze tests and have shown that the most economical and
reliable scoring method is the exact-word scoring method. Of the prominent
researches are those conducted by Taylor, Runkin, and Ruddell and Bormuth
(1976), and Brown (1983). Moreover, the use of the exact-word scoring
method yields to a slightly different result with native speakers that the
contextually acceptable scoring method, which is more tedious and requires
12
-
RC & LC Skills Devlpt. in al 11
more effort. (Oiler and Conrad, 1977, 1983, 1985)
Researches with non-native English learners however, such as those
of Oiler (1977), and Alderson (1983) show that the correlations between doze
and external validating criteria are significantly superior when the contextually
acceptable system of scoring is used. The most important arguement on the
use of the semantically acceptable scoring system was raised by Oiler in
defense of doze as a measure of intersentential comprehension. His findings
indicate that scoring for contextual appropriateness would probably have
augmented the difference across normal and scrambled conditions. "Further
since a contextually appropriate scoring could only be achieved by using the
original textual order as the criterial basis, it would have tended probably, to
create a priori bias against the scrambled condidtions." The exact-word scoring
method is too strict an application for the non-native English learners.
Furthermore the task of guessing the exact word placement is not necessarily
a language skill (Oiler, 1983). Other researches who have experimented with
the doze system on Fl 2 have often preferred scoring systems that give credit
for contextually acceptable answers.
Since the concern in the present study is to test the result of using
these methods for ESL Ss and EFL Ss, both the exact -word and the acceptable
scoring methods were 'KM The first method, in spite of its handicap for use
with non-native samples, was nevertheless carried out to provide a basis of
comparison as to which of the two systems of scoring will yield to higher
-
RC & LC Skills Devlpt in El, 12
correlations with the TOFEL listening test used. Assuming the arguements in
its favor- notwithstanding the difficulty of application - that the contextually
acceptable scoring method yields higher correlations and is rationally acceptable
with EL2 subjects, the second scoring method was applied with a native
speaker's judgement.
The Ss were instructed to use only one word to fill in the blanks. In
spite of this, however, there were instances when the respondents used more
than a one-word response. In the first scoring, these multiple responses were
counted incorrect. In the second scoring, however, these answers were judged
correct or incorrect on the basis of their acceptability in the context given in
the passage. Because of this and the re-scoring of the multiple-responses, the
second scoring method is said to have been less strict than the first method.
Test Analysis
To give an indication of the discriminative power of the doze test, the
data were analyzed by determining the means, standard deviation, mode and
the range. Table 1 summarizes the results of the first and second scoring of
the doze test plus the reliability coefficient scores.
-
RC & LC Skills Devlpt in az
Table 1 : Cloze Test Results with Reliability
13
Corefficien ts
FIRST SCORING SECOND SCORING
EFL ESL EFL ESL
No. of Data 116 114 116 114
Mean 19 28 25.16 40.45
Mode 18 30 29 39
Std. Dev. 19 15 8.76 5.32
Max. Score 36 40 43 49
Min. Score 5 20 6 28
Range 31 20 37 21
Reliability Estimates :Cron-Al .907 .801K-R 20 .907 .801
K-R 21 .854 .742
CLOZETEST RESULTS
The difference in the means obtained by the two groups after using
the second scoring system is much greater than that obtained after using the
first scoring method. The scores of the first group of Ss for the second
scoring, are on the whole, only slightly higher than their scores after the first
scoring been done. The mean scores obtained by this group (the EEL Ss)
in both scoring systems have been influenced by the very low base scores
which are 5 and 6. As far as the mode is concerned, however, majority ofthe
students in this group who scored below 50% in the first scoring - notice that
1 5
-
RC & LC Skills Devlpt. in EL, 14
the mode is 18 - performed above 50% in the second scoring - with the
mode of 29. The table likewise indicates the wide spread of some of the
scores of the EFL samples - with ranges of 31 and 37 - compared to the other
group - with a score range of 20 - 21.
The scores of the second group of Ss (ESL) are, on the whole, much
higher in the second scoring than in the first. Most of the students who
already scored above 50% in the first scoring obtained even higher when the
semantically acceptable scoring method was applied. Reliability estimates used
for the doze test were the Cronbach Aplha, Kuder-Richardson Formula 20
and Kuder-Richardson Formula 21. All the estimates gave .8 and .9 coefficients
for the Japanese Ss and slightly lower coefficients for the Filippo students.
1137ENING COMPREHENSION TEST RESULTS
Table 2 reports the means and standard deviation together with the
reliability estimates of the listening comprehension test used The results indicate
that in contrast to the doze test results, the spread of the scores of EFL Ss is
much narrower than that of ESL Ss. On the other hand, the raw scores of the
latter is much higher than those of the former, notice the 55.7 and 45.5 mean
scores respectively, with a lower STD in the ESL group (5.5) than in the
other (8.1)
-
RC & LC Skills Devlpt. in Ell 15
Table 2' Listening Comprehension Test Resultswith Reliability Estimats4
t Ft ESLNo. of Data 116 114
Mean 45.53 55.70
Mode 49 57
Std. Dev. 8.117 5.526
Max. Score 65 63
Min. Score 27 19
itangekeliability :
38 42
eron-Al .832 .805
IC-II 20 .832 .805
1K-R 21 .808 .746
For both tests, the performance of the ESL students was higher than
that of the EFL samples. This is indicated by the higher mean performance as
well as the smaller standard deviation of the former.
Significantly, one student from the EFL group got a perfect score
while the highest score in the second group was two points lower. Meanwhile,
the single score which best represents the performance of Group 1 is lower
than that of Group 2, with the performance modes of 49 and 57, with the
mean scores being 45 and 55, respectively. The reliability estimates were
obtained by using the Cronbach A 1pha, KR 20 and KR 21 formulas. The
coefficients obtained are in .8 range: for the first group of subjects; while for
the other group, they were found to %le within the .7 and .8 range.
1 '
-
RC & LC Skills Devlpt. in El2 16
On the whole, the test results obtained from the two groups of samples
indicate a low correlation, although positive. The slightly higher correlational
results of the ESL Ss indicate that the doze may not or does not necessarily
predict other language performance skills.
Fig. 1- 4 present the results of the correlational analysis done on the
doze and listening comprehension tests.
50
00
40r=0.502
Y, =1.301+0.383 X
30S.
§ 20 .' ".
UV) ..
eu X 104.1
0 10 20 3040 50 60
70
X=LISTENING COMPREHENSION TEST
Fig. 1: EFL =LCT and RCT (ExactWord Scoring
Method) St
9U
-
RC & LC Skills Devlpt. in az
5043
0U3
40.0 -tg
.9. 4)30
0, ty,
O.
o-n 20c.)
cts
euC-1 10u
CG
II
r=0.516
Y,=- 0.248 +0.572 X
17
10 20 30 4050 60 70
X=LISTENING COMPREHENSION TEST
Fig. 2: EFL =LCT and RCT (AcceptableScoring
Method) S,
040
.11-0
O204.1.5
di
E beocU
bary3
43 tico coca
c411.1
50
40
30
20
10
0
r=0.260
Y, =20.029 +0.172 X .
j
:.
e
10 20 30 40 50 60 70
X=LISTENING COMPREHENSION TEST1 CA
Fig. 3: ESL=LCT and RCT (Exact Word ScoringMethod) S,
-
1
RC & LC Skills Devlpt. in El2 18
-14
2
U1
c-5O N
."c5
a) .5
ocn.
4.)
be caz(til'
VUca'N 4CI
I
50
40
30
20
10
0
r=0.489
Y1=33.345+0.138 X
.
.:
t.I
I
4
10 20 30 4050 60 70
X =LISTENING COMPREHENSION TEST
Fig. 4: ESL =LCT and RCT (AcceptableScoring
Method) S2
Discussions of Results
Contrary to expectations, the results of the study yielded very low,
although positive correlations between the two tests. Comparatively, the results
of the tests of the EFL Ss gave a higher correlation than those ofESL Ss.
Furthermore, the acceptable scoring method (S 2) in Fig. 2 and Fig. 4 yielded
to a slightly higher correlation with the TOFEL test performance of both
groups than did the exact scoring method (S1) in Fig. 1 and Fig. 3, Finally,
the inter-group correlational difference with the application ofthe two scoring
methods varies and that such a difference is much higher in the second group
-
RC & LC Skills Devlo. ir. 1E12 19
(Fig. 3 and 4) than the difference in the scores of the first ,roup (Fig. 1 and
2). With the application of the semantically acceptable scoring method, a
higher correlation with an external criterion - such as the TOFEL - was
obtained It should be noted that the results were obtained by using the PPM
Correlation formula that makes use of data from raw scores.
Statistically, the data reported here are interesting; first in that it is
likely that the high schools students who were taken as samples comprised a
relatively homogenous group of individuals, and this curtailment of range,
specifically of age, may have resulted in spuriously low correlations.
Another fatter that might have greatly influenced the very low correlation
of the test results in the second group is the presence of "outliers". Fig 3 and
4 show the existence of the two greatly deviant scores which fall below the
lowest score in the listening comprehension test. The reason for the existence
of these "outliers" can best be studied through a careful case study, which
could not be included in this present study due to apparent reasons of space
and time. The clustered test scores may have resulted in the low correlation.
As regards doze test scoring methods, a tentative conclusion that can
be drawn from the data of this research is that between the exact and contextually
acceptable scoring systems, the latter yields higher correlations with other
tests than the former. This confirms the results of some previous researches,
V
(7-
-
RC & LC Skills Devlpt. in El, 20
especially those of Darnell, et al (1976) and Vollmer and Sang (1983) that the
contextually acceptable scoring method is superior in terms of correlational
results when applied to samples of non-native speakers.
The higher mean scores and higher correlation results with the listening
comprehension results of both groups of samples when the seanantically-
acceptable scoring system was used support the claim that it is the rationally
justifiable with subjects to whom English is a second or a foreign language.
The difference in the performance of the two groups of non-native.
English learners present striking insights not only into the relationship between
listening comprehension and reading but also tena to LIM.t-cr:t.izt-.-..;:ze.crits'
exposure to English as a second or foreign language. A number of factors
may be thought of, the most influential of which may be the degree to which
they are exposed to and use English in the school, at home and in the
environment
Although it is not the aim of this paper to explore the complex process
of reading in an SL or FL setting, it is interesting to look into how the skills
of reading and listening to English as FL or SL is organized and how they
develop or are developed. But in answering this question, one is really
asking, as Carroll suggests, whether these skills are attained at the same rate,
or attained separately, at different rates (1983). It is possible to trace the 3=31
rate of skill development by subjecting the samples to certain observational
-
RC & LC Skills Devlpt. in Ell 21
criterion over a period of time and using the refined factor analysis. A high
correlation between a pair of skill variables would indicate that to the extent
the learner has attained one of these skills, he is most likely to have attained
the other skill at about the same relative extent. The findings of this research
support this claim. The comparatively high correlation between the doze and
listening test results of the Japanese students indicates that exposure to English
as a foreign language has made possible the fact that the language skills have
been tapped and used at an equally same degree at this stage.
This implies that the cognitive skills learned in decoding the messages
in both the audio and written tests have advanced to a certain extent that such
advancement in the listening skills is accompanied by advancement in the
other skill, or vice-versa (Carroll, 1983). But what does this indicate in
relation to the samples to whom English is a second language?
For the second group of population, with English as a medium of
instruction in school, and for most of whom English is the major language at
home and in the environment, the low correlational test results may indicate a
fairly unequal development of the receptive skills of audio-visual comprehension
such that the skills in listening comprehension have advanced to a certain
degree while the other skills of reading comprehension have not. It may
happen that in a second language environment, the students are exposed more
to aural English, the language they first come in contact with, and more
because it is a common language in mass media, and even more when it is
-
RC & LC Skills Devlpt in Ell 22
used as a medium of instruction. Or it may be the reverse; it is hard to
determine at this point which skill has been developed more and which has
advanced less. The close test results can not point out to this. The findings
simply indicate that there is a gap in the level of skills advancement
But just how exactly these language skills develop in order is unknown
in SLlearning (Verhoeven, 1990) and Nagle and Sanders (1966). The extensive
discussions of James and Sticht (1982) on the relationship between listening
and reading provide helpful insights on this area of research. Although the
article discus 7s mainly issues on reading and listening in relation to Ll, it
nonetheless throws light on this study, assuming a Unitary Competence and
A General Language Proficiency Theory. Besides, as Oiler (1977) has earlier
pointed out in relation to the doze, reading and listening tasks require the FL
and SL learners to perform a task which is not unlikely to what native speakers
do in receiving and sending informatice. Furthermore, studies on ESL2 learners
suggest that L2 acquisition in naturalistic context has much in common with
first language acquisition. (Elley and Mangubhai, 19E3).
The most important arguement raised in the article of James and Sticht
is that the development of the skill of auding is a pre- requisite to the development
of the skill in reading. The reading potential concept developed by James and
Sticht indicates that auding and reading performance will be more highly
correlated after some skills of learning to read has developed. This may have
been the case with the EFL sti lents taken as samples. However, this does not
mean that the ESL learners have not learned to read better than the al, - quite
P
-
RC & LC Skills Devlpt in 132 23
the contrary, as seen in higher reading test results. The higher mean scores
and lower Sir in the listening test- compared to the other group and to their
own (ESL) reading scores is not unexpected considering the fact that English
is a near-native language; it is "common", interchangably used with - or even
more than - the native tongue. This being the case, they would have had the
potential for higher auding skill. And according to James and Sticht's claim
on the transfer effect, pupils highly skilled in auding would have (greater)
potential to become highoy skilled in reading, which is proven by higher
reading scores.
But the correlationally lower test results of the ESL samples support
another issue raised by the article that although exposure to spoken language
is necessary for reading to take place, auding-to-reading transfer may not
necessarily occur in the auding, then reading order. It may have been the
reverse. Congruent with this arguement, the results with the EFL Ss indicate
that a reading first, then, auding transfer may have been passible because the
age of the Ss (3rd year JHS) mean that they may have had the time to learn the
peculiarities of written English more -which reflects strongly the emphasis on
grammar and translation method very common in Japanese schools.
On the whole, the relatively low correlation in the test results of both
the EFL and ESL students lead to a question on the unitary competence
hypothesis postulated by Oiler. "If the components of language competence
-
RC & LC Skills Devlpt in 22 24
function more or less similarly in any language-bared task, high correlations
wou'd be expected between language test scores of all sorts, whatever modality,
whatever format."(011er, 1983) Given the limited conditions and the limited
number of samples for this study, what can be tentatively stated is that the
different language skills or listening and reading are learned together and
developed in different degrees according to whether exposure to English is as
a FL or as an ESL, and yet there is a general tendency for some language
skills to be developed and specialized to some degree than the others and can
be measured and recognized as distinct from other language skills. Thus the
results of a doze test may be, but not the only predictor of the listening
comprehension ability of English as the foreign or second language learners.
As regards doze test scoring methods, another tentative conclusion
that can be drawn from the data of this research is that between the exact and
contextually acceptable scoring systems, the latter yields higher correlations
with other tests than the former. This confirms the results of some previous
researches (Alderson, 1983 and Bachman 1982,1985) that the contextually
acceptable scoring method is superior in terms of correlational results -
especially with L2 samples - compared to other scoring systems.
Because of the very limited time involved in this study, the relatively
small number of population and the limitations of only one doze test and one
listening comprehension test used, the findings and conclusions arrived at are
only inclusive of the present data. Other important areas of concern have not
-
RC & LC Skills Devipt in Ell 25
been addressed in this study, and are open to further research.
Fust, a response frequency analysis need be carried out for each item
of the doze test and / or listening comprehension test as completed by each
group of subjects. The main question to be addressed here is whether or not
the responses of the EFL students would be significantly different in kind as
well as in number from those of the ESIstudents.
Second, a further experimentation with another improved doze testof
a different difficculty level need to be conducted and administered to a bigger,
more heterogenous population than those used in this experiment, and that the
test be further correlated with other external validating criteria. Why the second
group of Ss scored lower in spite of the use of the TOFEL exam in listening
need further investigation.
Third and closely related to the second, the low reading test results
point out to a need to look into the reading problems that go with a2 study.
The complex mechanism involved in the acquisition of reading in EL2discussed
in Verhoeven (1990) runs along parallel lines with the discussions of James
and Sticht (1982) that emphasize that children should be helped to build up
oral skills in EL2 first before reading instruction is started. In the case of the
samples used in the present study, instructional reading materials arepresented
simultaneously with audio instruction, although environmental exposure to
auding starts earlier in the Esi, Ss than in the EFL Ss.
9 7
-
RC & LC Skills Devlpt. in El, 26
Then too, if the low correlation between the tests indicate a gap in the
development of the language skills of the non-native English students, a more
systematic search be started for factors related to increases in English language
proficiency at various stages of ESL and EFL studyunder diverse cirwristances
of language learning as those of Japan and the Philippines. (OrMally and
Chamot, 1985) and Vollmer and Sang (1983). This experiment has identified
a gap in the development of the language skills and leaves room for dr next
task which is for the researches and for those in the field to reduce the
obvious disparities in competence levels as a resultof such gap.
9. 3
-
RC & LC Skills Devlpt in El2 27
REFERENCES
Alderson, J.C. (1980). Native and Non-native Speaker Performance on doze
test, Language Learning, 30, 59-76.
Alderson, J. C (1983). The Claw Proficiency and Proficiency in English as
a Foreign Language. In John Oiler (Ed), Issues in Language Testing
Research (pp. 205- 217). Rowley, MA: Newbury House.
Anderson, J. (1977). A Technique for Measuring Reading Comprehension ,
Language Learning , 21, 178 -181.
Bachman, L (1982). The Trait Structure of Cloze Test Scores, /P..,OL
Quarterly, 16, 61-67.
Bachman, L (1985). Performance on Cloze Tests with Fixed-Ratio and
Rational Deletions, TFSOL Quarterly, 19, 535 - 551.
Belasco, S. (1965) Nucleation and Audio-lingual Approach, Modem
Language Journal., 48, 217 - 223.
Belasco, S. (1978). Feasibility of Learning English in a SL Unicultural
Situation, In Pimsleur Psychology of Second Language Learning. (1 -
10 ). Cambridge: University Press.
Brown, J. (1983). A Closer Look at Cloze. In Oiler, J. (Ed.), Issues in
Language Testing Research (pp. 237 - 250). Rowley, MA: Newbury
House.
Carroll, J.B. (1977). Development Parameters of Reading Comprehension.
In J.T. Guthrie'(Ed), Cognition, CUMCUilerl and comprehension (pp
-
3
RC & LC Sails Devlpt in El, 28
11 -15). Newark, DE:Intemational Reading Association.
Carroll, J.B. (1983). Psychometric 'Theory. in John Oiler (Ed), Issues in
Language Testing Research: (pp. 80 -107): Newbury House.
Chapelle, A. and Abraham, R (1990). Cloze Method: what difference does it
make? Language Testing, 7, 121- 146.
Chastain, K (1976). Developing Second Language Skills. Boston:
Houghton Miffin.
Chavez-Oiler, Chihara, T., & Oiler, J. (1985). When are Cloze Items
Sensitive to Constraints Across Sentences? Language Learning, 35, 181
- 202.
Chihara, T., Oiler, J.W., K-Weaver, & Chavez-Oiler, (1977). Are Cloze
Items Sensitive to Constraints Au ubs Sentences? Language Learning,
27, 63 - 73.
Dale, E & Chall, J.S A Formula for Predicting Readability. Educational
Research Bulletin, 1948, 27,11 -20, 37 - 54.
Durrell, D.D. (1%9). Listening Comprehension VS Reading
Comprehension. Journal of Reading, 12, 455-460.
Elley, W. & Mangubhai, F. (19E3). The Impact of Reading on Second
Language Learning, Reading Research Quarterly, 19, 53 - 67.
Koike, I. Ed. (1978). The Teaching of English in Japan. Tokyo: Eichosa
Publications.
Lado, R (1980). Language Testing. (pp. 315 - 333). London: Longman
Press.
Lado, R (1986). Analysis of Native Speaker Performance in Cloze Tests.
3 ti
-
RC & LC Skills Devlpt. in El, 29
Language Testing,3,130 -153.
Markham, P. (1985). The Rational Deletion Cloze and Global
Comprehension in Germ= Language Learning, 35, 423 - 430.
Nagle, S. & Sanders, S. Comprehension Theory and Second Language
Pedagogy. TESOL Quarterly, 20, 9 - 24.
Oiler, J. Ed. (1983). Issues in Language Testing Research. (205 - 217).
Rowley. MA: Newbury House.
Oiler, J.W. & Hinofitis, F.D. (1980). Two Mutually Exclusive Hypotheses
about Second Language Ability. In Oiler and Perkins (Eds.), Research
in Language Testing, Rowley:MA: Newbury House.
Oiler, J. (1979). Pscychoiogy of Language and Contrastive Linguistics.
Foreign Language Journal, 12, 299-309.
Oiler, J. (1972). Assessing Competence in ESL: Reading. TESOL Quarterly,
6, 321- 323.
Chamot, A. et al. (1985) Learning Strategies Used by Beginning
and Intermediate ESL Students. Language Learning, 35, 21-46.
Perkins, K (1983). Semantic Constructivity in ESL Reading
Comprehension. TESOL Quarterly, 17,19 - 27.
Shanahan, T., Kamil, M. & Tobin, A.W. (1932) Cloze as Measure of
Intersentential Comprehension. Reading Research Quarterly, 17,
229-255.
Shikhani, M. & Hanania, E (1936). Interrelationships Among Three Tests of
Language Proficiency. TESOL Quarterly, 20, 97 - 122.
Sticht, T. and Jame's, J. (1984). Listening and Reading. In P. Pc..arson, R
31
-
tiR
RC & LC Skills DevIpt in EL,. 30
Barr, M. Kamil &Mosenthal, P. (Frig ), Handbook of Reading
Research New York Longman. (2S (3-318).
Towsend DJ., Carrithers, C., & Bever, T.G. (1987) Listening and
reading process in college-and middle school-age readers. In R
Harosvitz & Si. Samuels (Eds.), Comprehending Oral and Written
Language ( 217 - 242). New York Academic Press.
Verhoeven, T. (1990). Acquisition of L2 reading,Reading Research
Quarterly, 25, 91 - 112.
Vollmer, FL & Sang, F. (1983) Competing Hypotheses about Second
Language Ability. In John Oiler (Ed.), Issues in Language Testing
Research , (pp. 29 - 79). Rowley MA: Newbury House.
Winitz, FL Ed. (1981) Native Language and Foreign Language Acquisition.
New York Academy of Science Press.
32